Axom MCP Server
Provides persistent SQLite-based memory and unified tool abstraction for AI agents to support long-term context and complex tool chaining. It enables automated code analysis, file operations, and environment discovery through a standardized interface.
README
Axom MCP Server
Axom is a Model Context Protocol (MCP) server that provides persistent memory, tool abstraction, and chain-of-thought for AI agents.
Core Features
- Persistent Memory: Store and retrieve context across sessions using the Axom (SQLite) database.
- Tool Abstraction: Unified interface for memory, execution, analysis, discovery, and transformation.
- Chain Reactions: Execute tool sequences where outputs feed into the next step.
- AI-Powered Classification: Automatically categorizes memories by type and importance.
Quick Start
Axom runs as stdio MCP - your IDE spawns it automatically. No manual server startup needed.
Prerequisites
- Python 3.11+
- SQLite (included with Python)
- Git
Installation
Linux / macOS / WSL / Windows (Native)
Requires Git Bash, PowerShell, or a make provider.
git clone https://github.com/PugzUI/axom-mcp.git
cd axom-mcp
make install
What make install does:
- Installs Python dependencies.
- Installs Axom in editable mode (
pip install -e .). - Creates
.envfrom.env.example. - Configures all detected agents (Cursor, Trae, etc.).
- Installs Axom rules and skills for each agent.
Verification
make test # Run tests
Post-Install Check
Verify the local command is available:
command -v axom-mcp
Verify MCP initialization against the installed server:
python3 - <<'PY'
import asyncio
from mcp import ClientSession
from mcp.client.stdio import stdio_client, StdioServerParameters
async def main():
params = StdioServerParameters(command="axom-mcp", args=[])
async with stdio_client(params) as (read_stream, write_stream):
async with ClientSession(read_stream, write_stream) as session:
await session.initialize()
tools = await session.list_tools()
print("tool_count:", len(tools.tools))
asyncio.run(main())
PY
Client Configuration
make install automatically configures MCP for detected agents. The installer uses the best available command:
axom-mcp(if in PATH)axom(if in PATH)python -m axom_mcp.server(fallback)
See docs/agents/INDEX.md for detailed agent configuration.
For Zed, ~/.config/zed/settings.json should contain:
"context_servers": {
"axom": {
"command": "axom-mcp",
"args": []
}
}
Tools
Axom provides five core MCP tools:
axom_mcp_memory: Store and retrieve persistent context.axom_mcp_exec: File operations and shell commands with chaining.axom_mcp_analyze: Code analysis and debugging.axom_mcp_discover: Map environment and capabilities.axom_mcp_transform: Convert data between formats.
Documentation
- Architecture - System design and data flow.
- Tool Reference - Detailed tool parameters.
- Agent Guide - How to use Axom with AI agents.
- Troubleshooting - Common issues and fixes.
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。